Massachusetts Institute of Technology Lincoln Laboratory an Optimum Interference Detector for Dabs Monopulse Data Editing

1. INTRODUCTION The ability of a monopulse processor to determine the angular direction of an incident signal is 1imited not only by the inherent front-end receiver noi se but also by the effect of interfering signals. If there are other time coincident signals present that interfere with the signal whose direction is to be estimated then McAulay [1] has shown that bias effects occur that can seriously degrade the quality of the estimate. It becomes important therefore, to know when such interference is present so that low confidence can be assigned to the associated azimuth estimate. Sherman [2] has observed that when interfering signals are present the outputs of the monopul se sum and difference beams become incoherent. He proposes to use the quadrature information to resolve the target and interference signals. Our one the approach is to develop an interference flag that indicates when more than signal is present in the receiver channels. Depending on the application, flag would be used to assign a low confidence to the associated angle estimates or to delete the angle the test for interference as Ratio test [3] and following estimate altogether. In Section 11, we formulate a hypothesis test. Using the General ized Likelihood the analysis of Hofstetter and DeLong [4], we obtain the optimum interference detection statistic. shown that the interference statistic has the Rician distribution, Exact evaluation of the false alarm and detection probabilities becomes intractable and use is made of the Gaussian approximation to the Rician variate. Numerical 1 results for some typical cases of interest are given and it is shown that the results depend strongly on the relative phase between the target and interfering signals, but that good overal 1 performance can be obtained. Conditions under which the Gaussian approximation is valid are given and are shown to hold for the cases studied. We shal 1 restrict our attention to ampl itude-comparison monopul se processing that is performed on a sampled-data basis. Assuming mixer preamplifiers at the output of each antenna beam channel, the received signal samples in the absence of interference are modeled by jw~ yi=ASe where yi refer to the complex output of ith antenna beam channel; AS.qS,~S,CXS are the amplitude, phase, azimuth, and elevation of the target signal; Gi(o) is the .th antenna patterns of the 1 antenna beam which may be complex in general; ni represents zero mean Gaussian noise samples due to …